1. Field of the Invention
The present invention relates to the waveform shaping equipment and waveform-shaping method for generating bandlimited signals, and for preventing band spread at the head and trail at the edge of burst when burst-like data string is transmitted in the data transmission in which data is transmitted in the form of packet.
2. Related Art of the Invention
In the radio communication, etc., when a packet comprising transmission data is transmitted, it is necessary to limit the bandwidth (bandlimitation) to prevent adjacent channel interference for effective utilization of frequency. For bandlimitation of signals, it is common to limit the bandwidth with respect to the signal waveform of the baseband. Two systems are available for band-limiting the baseband signal waveform: an analog system using analog filter and a digital system by digital signal processing. One of the digital systems is the method to shape waveform by reading out and concatenating the baseband signal waveform previously band-limited by calculation from the memory table such as ROM and the like (for example, IEEE transactions on Communications, COM-vol. 25, No. 10, Pages 1243-1244). When the waveform shaping method using this memory table system is used, the ideal filter frequency response can be more accurately realized than analog system waveform shaping method, and the shaped waveform can be changed only by rewriting the memory contents, achieving high versatility. It is also suited for the VLSI technique and can be comparatively downsized.
Referring now to the drawings, the conventional waveform shaping equipment using the above-mentioned method is described with special emphasis placed on the readout principle of shaped waveform and hardware configuration of the waveform shaping equipment.
FIG. 1 shows input data to the waveform shaping equipment. D(1), D(2) . . . , D(k), . . . D(n) show transmission data and X shows the data other than the transmission data,which does not have any information. Each data is successively read into the waveform shaping equipment at every time interval T.
FIG. 2a shows the data pattern comprising each input data of FIG. 1. The data pattern is used to specify part of the address for reading out waveform after bandlimitation from the memory table. In this section, to simplify, description is made supposing that there is an intersymbol interference which has 3 symbols time and the data pattern length is 3 symbols. A(1), A(2), and A(3) show a time slot, respectively. Let the time slot A(2) in each data pattern be the present time slot. Then, time slots A(1), A(3) affect the present time slot A(2) by intersymbol interference. Each data pattern (p(1), p(2), p(3), p(4), data, respectively, and the data pattern p(1) comprises the data (D(1), X, X), the data pattern p(2) comprises the data (D(2), D(1), X), the data pattern p(3) comprises the data (D(3), D(2), D(1)), the data pattern p(4) comprises the data (D(4), D(3), D(2)), the data pattern p(n) comprises the data (D(n), D(n-1), D(n-2)), the data pattern p(n+1) comprises the data (X, D(n), D(n-1)), and the data pattern p(n+2) comprises the data (X, X, D(n)).
FIG. 3 shows the case when the data pattern corresponding to the present time slot which varies at every time interval T is extracted.
FIG. 2b shows baseband waveform after bandlimitation, which is generated when the waveform is read out from the memory table successively at every 1 symbol time T by the data pattern shown in FIG. 3. That is, the waveform w(3) which has 1 symbol time is generated by the data pattern p(3), the waveform w(4) equivalent to 1 symbol time is generated by the data pattern p(4), and the waveform w(n) equivalent to 1 symbol time is generated by the data pattern p(n). Because in the data patterns p(l), p(2), p(n+1), and p(n+2) indefinite data X with no information is contained, it is designed to output the 0-level waveform as the waveform for w(1), w(2), w(n+1), and w(n+2) at the time corresponding to data patterns p(1), p(2), p(n+1), and p(n+2).
FIG. 4 shows one example of a block diagram showing the hardware configuration of conventional waveform shaping equipment. In FIG. 4, S3 denotes a shift register, C3 a counter, M3 a memory table, D3 a D/A converter, and L3 a low-pass filter. dt3 denotes a dta string, co3 a counter output, so3 a shift register output, mo3 a memory output, wd3 a continuous waveform after D/A conversion wl3 a shaped waveform after smoothing. In general, let the data string dt3 be the data string of 2 M value (M: natural number), 1 symbol is M bits and the shift register 101 is made up of M bits.times.3 stages. Therefore, the output from each stage becomes M bit each, respectively. For simplification, description will be made assuming that the shift register handles M=1, that is,binary data.
The shift register S3 accumulates data for latest 3 bits of the data string dt3, and while taking in 1-bit data from the data string dt3 at every 1 symbol time and shifting, it outputs 3-bit data pattern so3 in parallel. The memory table M3 is a ROM which stores waveform data for one symbol time with the effects of intersymbol interference taken into account by prior calculation. That is, it stores waveform data for all the patterns which the total of 3 bits comprising the symbol to be transmitted and symbols before and after can take. Now, let the waveform data for one symbol time comprise 8 samples. The counter C3 is a 3-bit counter, which counts up 8 times in one symbol time and repeats operation with one symbol time as one cycle. The memory table M3 designates a total of 6 bits as an address, which comprises 3-bit data pattern so3, an output of each stage of the shift register S3, and 3-bit output co3 of the counter C3 which represents the location in one symbol time, retrieves the waveform data at each time corresponding to the data pattern to be transmitted, and outputs the memory output mo3. The memory output mo3 is converted to continuous waveform wd3 at the D/A converter D3 and after smoothed at the lowpass filter L3, it become shaped waveform wl3.
Next discussion will be made on the method for generating baseband signals after bandlimitation in the QPSK using this method. FIG. 5a shows data of the in-phase axis and quadrature axis extracted at every time slot from the transmission data string in the QPSK. Expressing this as a transition state for each time slot on the signal space produces FIG. 6. In FIG. 6, each signal point transitions at each time slot and the locus on the time axis of the orthogonal projection cast on the in-phase axis and quadrature axis of the coordinates of transitioning signal point represents the baseband signal waveforms of the in-phase axis and quadrature axis. FIG. 5b shows the baseband signal waveform of the in-phase axis and the quadrature axis corresponding to the in-phase axis and quadrature axis data shown in FIG. 5a before bandlimitation. When the baseband signal waveform of the in-phase axis and quadrature axis shown in FIG. 5b are band-limited with the intersymbol interference of the data pattern length taken into account, the baseband signal waveform after bandlimitation as shown in FIG. 5c can be obtained. The in-phase axis signal waveform and the quadrature axis signal waveform make the H level of waveform correspond to the data value "0" and the L level of waveform to "1" as shown in FIG. 5b and 5c. In the case of the QPSK, since the baseband signal waveform of the in-phase axis is determined by the in-phase component of the coordinates of each signal point and that of the quadrature axis by the quadrature component, the data patterns of the in-phase axis and the quadrature axis can be obtained separately from the in-phase component and the quadrature component in the time slot. In addition, because the baseband signal waveforms for the same data pattern of the in-phase and quadrature axes become identical, the waveform data necessary for shaping baseband signal waveforms of in-phase and quadrature axes can be used in common. Consequently, the storage capacity can be reduced as shown in, for example, the Japanese Patent Application Laid Open No. 1-317090.
FIG. 7 is a block diagram of waveform shaping equipment for the QPSK by the above-mentioned conventional method. In FIG. 7, C6 is a clock generation circuit, DV6 a 1/2frequency divider, DP6 a 2-bit shift register, SR6I a d-bit shift register, SR6Q a d-bit shift register, C06 a n-bit counter, DS6 a data selector, M6 a L-bit output memory table, SR6 a L-stage 2-bit shift register, FF6I a flip-flop, FF6Qa flip-flop, PI6 a .pi.-phase shift circuit, D6I a D/A converter, D6Q a D/A converter, L6I a low-pass filter, and L6Q a low-pass filter. ck6 is a system clock, ckd6 a divided clock, ckp6 a .pi.-phase shift clock, dt6 a data string, dt6I an in-phase axis input data, dt6Q a quadrature axis input data, so6I an output of shift register SRI6, so6Q an output of shift register SRQ6, co6 a counter output, mo6 a memory output, wd6I an output of D/A converter D6I, wd6Q an output of D/A converter D6Q, w6I shaped continuous waveform of the in-phase axis, and w6Q shaped continuous waveform of the quadrature axis. For simplification, description will be made when the equipment treats the case in which d=3, n=2, and L=3.
The shift register DP6 takes the data string dt6 at every 1 clock while shifting, retains the latest 2-bit data, and outputs by allotting one bit each to the shift register SR6I and the shift register SR6Q. The shift register SR6I and the shift register SR6Q take in the output of shift register DP6 one bit at a time as in-phase axis data dt6I and quadrature axis data dt6Q while shifting every 2 clocks by the divided clock ckd6, hold the latest 3-bit data, respectively, and output the shift register output so6I and shift register output so6Q in parallel as a 3-bit data pattern for the in-phase axis and quadrature axis, respectively. Now, the waveform data for 1 symbol time comprises four samples, and using the 2-bit counter CO6 whose 1 symbol time is 1 cycle, part of the address of waveform data to be read out within one symbol time is specified based on the counter output co6. The memory table M6 which has a 3-bit output is a ROM which stores waveform data for 1 symbol time with the effects of intersymbol interference taken into account by prior calculation. That is, the ROM stores waveform data quantized by 3 bits for all patterns which can be taken by the total of 3 bits comprising the symbol to be transmitted and those before and after it. The shift register output so6I and shift register output so6Q which are data patterns of the in-phase axis and quadrature axis time-share the waveform data in the memory table M6 by being selected by the data selector DS6 alternately and becoming part of the address. The 3-bit memory output mo6 read out alternately from the data pattern of the in-phase axis and quadrature axis, respectively, are allotted to the flip-flop FF6I and flip-flop FF6Q by the 2-bit 3-stage shift register SR6, which shifts every 1 clock, and are taken in simultaneously to the flip-flow FF6I and flip-flop FF6Q by the clock timing ckp6 generated by the .pi.-phase shift circuit PI6. In addition, the outputs of the flip-flop FF6I and flip-flop FF6Q are converted to the analog waveform wd6I of the in-phase axis and the analog waveform wd6Q of the quadrature axis via the D/A converter D6I and D/A converter D6Q, and after smoothed at the low-pass filter L6I and the low-pass filter L6Q, they are formed into the shaped waveform w6I, which is the baseband signal of the in-phase axis, and the shaped waveform w6Q, which is the baseband signal of the quadrature axis. FIG. 8 shows operation timing of each section of the equipment. In the case of QPSK, it has been possible to reduce the memory capacity to one half, which is required to store waveform data by the waveform shaping equipment shown in FIG. 7.
However, in the case of quadrature modulation in which the in-phase axis and quadrature axis baseband signal waveform to be read out must be determined based on both inphase and quadrature components of the coordinates of the signal point corresponding to the transmission data, for example, in the .pi./4 shift QPSK and in the PSK-VP (phase shift keying with varied phase) system described in Pages 412-419 of the proceedings of the 40th IEEE Vehicular Technology Conference), it is impossible to extract the in-phase component and quadrature component data independently at the in-phase axis and quadrature axis for each time slot and form a data pattern.
This will now be described by way of example in the case of the .pi./4 shift QPSK. FIG. 9a shows the transmission data for each time slot in the .pi./4 shift QPSK. The signal point corresponding to the transmission data of each time slot takes the transition state as shown in FIG. 10 on the signal space. In FIG. 10, A9e shows the quadrature coordinate axis corresponding to the even-number time slot, and A9o shows the quadrature coordinate axis corresponding to the odd-number time slot when the quadrature coordinate axis A9e is rotated 45.degree.. In FIG. 10, the signal point transitions every even-number time slot and odd-number time slot with the quadrature coordinate axis varied, and the locus of the orthogonal projection which the coordinates of the transitioning signal point cast on the in-phase axis and quadrature axis on the time axis represents the baseband signal waveform of the in-phase axis and that of the quadrature axis. FIG. 9b shows the baseband signal waveforms of the in-phase and quadrature axes before bandlimitation, which correspond to the transmission data shown in FIG. 9a. Band-limiting to the in-phase and quadrature axes baseband signal waveforms shown in FIG. 9b, produces the intersymbol interference and can provide the baseband signal waveform after the bandlimitation as shown in FIG. 9c. In the case of .pi./4 shift QPSK, in-phase and quadrature baseband signal waveforms corresponding to the transmission data depend on both in-phase and quadrature components of the coordinates of the signal point. That is, this is also apparent from the fact that the in-phase baseband signal waveforms read out at the time slots t4 and t6 shown in FIG. 9b differ because in the even-number time slots t4 and t6 of FIG. 9a, each in-phase component takes the same data value "1," but each quadrature component differs. This means that for reading out waveform for both in-phase and quadrature axes, it is necessary to use as part of the address (1) the data pattern which comprises double bit number including both in-phase and quadrature components for each time slot and (2) the signal which selects quadrature coordinate axis either A9e or A9o respectively. In the case of .pi./4 shift QPSK, the same data pattern is used for the in-phase and quadrature axes, but because the baseband signal after bandlimitation of in-phase axis and quadrature axis to be read out for the same data pattern differs, respectively, it is unable to take a configuration to time-share the memory table as shown in FIG. 6 and it must be designed to store in separate memory tables, respectively, the baseband signal waveforms in all cases with the effects of inter-symbol interference from several symbols for the in-phase and quadrature axes taken into account.
However, in the configuration in which all the baseband signal waveforms for the above in-phase and quadrature data patterns are stored in separate memory tables, respectively, suppose that the number of symbols which have effects on the intersymbol interference is d, the number of samples in one symbol is n, and the quantization bit number of waveform data is L, the memory capacity required to retain the waveform data becomes 2.times.2 (3d).times.L.times.n bits in the case of the .pi./4 QPSK and 2.times.2 (2d).times.L.times.n bits in the case of the QPSK-VP, creating a problem that the memory capacity greatly increases as compared to 2 d.times.L.times.n bits of the QPSK.
When burst-like data strings are transmitted by each of the above systems, for example, in the case of FIG. 2b, abrupt rise and fall of waveform occur at the burst edge at the head and the trail of the data string at non-continuous points qb and qc, causing the spectrum to spread and the band to expand. Consequently, it becomes also necessary to shape the waveform smoothly at the burst edge. Conventionally, in waveform shaping at this kind of the burst edge, for example, as described in the Japanese Patent Application Laid Open No. 4-58622, waveform shaping is generally carried out by installing a variable gain amplifier or a variable attenuator at the portion where the waveform is amplified and varying the gain or attenuation rate smoothly at the start and at the end of data string.
The conventional burst waveform shaping equipment using the above-mentioned method will now be described with reference to the accompanying drawings.
FIG. 11 is a block diagram illustrating a conventional burst waveform shaping equipment and FIG. 12 is a diagram showing waveform at each section of the burst waveform shaping equipment in FIG. 11. In FIG. 11, WG10 is a continuous waveform shaping means, VA10 a variable gain amplifier, and CS10 a gain control signal generating means. In FIGS. 11 and 12, dtlO is a data string, wo10 a shaped continuous waveform, vo10 an output signal, and co10 a gain control signal.
The data string dt10 is the burst-like data composed by arranging preamble pre, information data info, and postamble post in that order as shown in FIG. 12. Of these, info is the data string to be transmitted and pre and post are data strings which do not carry information. The contents of pre and post may be optional but at this point, as an example, the 0101 4-bit data string is assumed for both.
The continuous waveform shaping means WG10 is a circuit similar to the above-mentioned waveform shaping equipment and outputs shaped continuous waveform wo10 which is shaped to have smooth waveform at the data continuing portion.
At first, the gain control waveform generating means CS10 generates the gain control signal co10 and controls the gain of the variable gain amplifier VA10. In this event, when the gain control signal co10 is zero, the gain of the variable gain amplifier VA10 is zero and as the gain control signal co10 increases, the gain also increases. The gain control signal co10 is zero in the period without data, smoothly increases from zero to a specified level in the period of preamble, holds the specified level during the period of information data, and smoothly decreases to zero from the specified level in the period of postamble. Consequently, the output signal vo10 outputted by the variable gain amplifier VA10 has a zero amplitude during the period without any data string to be transmitted, smoothly increases the amplitude in the preamble interval before the data string to be transmitted starts, and smoothly decreases the amplitude in the postamble interval when the data string to be transmitted ends.
With the above mentioned operation, the output signal vo10 is obtained by multiplying the output of the continuous waveform shaping means by the gain waveform of the variable gain amplifier, and because the waveform smoothly varies even at the head and the trail of the data string, the spread of spectrum during transmission of the burst-like data can be prevented.
When carrier transmission is carried out, it is common to generate baseband waveforms as the shaped continuous waveform wo10 and to carry out burst shaping using a variable gain amplifier at the high-frequency amplified portion after the carrier is modulated with wo10.
However, with the above-mentioned configuration, a variable gain amplifier for burst shaping and a gain control means are required in addition to the continuous waveform shaping means. Furthermore, to prevent spectrum spread, it is necessary to hold the gain change adequately gentle; this requires at least several symbols for the preamble and postamble lengths during the period when the gain is varied.